Skip to content

omamobile-omarchy 0.1.39: select 4.0.6 and 4.5 patch families - #471

Merged
derekross merged 5 commits into
derekross:mainfrom
marselnenaj:codex/qualified-45-runtime
Nov 10, 2026
Merged

derekross merged 5 commits into
derekross:mainfrom
marselnenaj:codex/qualified-45-runtime

Conversation

@marselnenaj

@marselnenaj marselnenaj commented Nov 7, 2026 •

Copy link
Copy Markdown
Contributor

Select the mobile patch family from the installed Omarchy version: 4.0.6 uses legacy-4.0.6/, while 4.5 uses patches/. AGENTS, README and RELEASES explain where fixes for each source family belong. Unsupported versions are refused.

Rebased onto main at b5207b86. Version 0.1.39 follows the maintainer's allocation. All 29 legacy patches retain main's exact bytes and notices, including #420's 0029 double-tap-to-wake patch and jkjoplin's merged #431. The 0029 patch is moved into the legacy family; the 4.5 family remains the same 26 patches as the previously tested PR head. The #431 4.5 port remains omitted. Omarchy's original copyright and MIT terms are preserved, and the mobile fixture additions retain their attribution.

The runtime updater, migration holds already merged through #486, and repository package pins are unchanged from main. This branch contains no shared #469 commit and does not change the normal update flow.

Tests use packaged lock sources and repository fixtures. Sleep-lock tests select their family from the actual archive version, without following the build host's absolute bin symlinks. The private-lock rendering override is removed. Node and QML tools remain optional and explicitly report skips when unavailable.

A valid notification key that is already absent returns ok without mutating state; malformed keys still return none.

Validation of this rebased head:

  • test/all passed separately against the cached real aarch64 Omarchy 4.0.6-1 and 4.5.0-3 archives.
  • Both real-family apply/verify/restore rehearsals passed, including idempotence, external-edit refusal, packaged PAM preservation and sleep-lock behavior.
  • All 15 touched legacy QML files and 16 touched 4.5 QML files parsed. The legacy lock-glance Qt suite passed 56 tests with no failures or skips; the available Node checks also ran.
  • All 29 legacy patch hashes equal current main's former flat series; all 26 4.5 patch hashes equal the previous PR head.
  • Commit-range policy checks passed. Non-patch source whitespace checks passed; existing patch payload bytes were preserved.

The earlier maintainer-reported cheetah 4.5 phone rehearsal belongs to the previous head. This local revision was not installed on a phone. The requested fresh 4.0.6 phone check remains with the maintainer, and these package rehearsals do not claim that all pending Omarchy migrations ran. The separate 0017 animation follow-up is not bundled here.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj, this is a big piece of work, and the series port is very good. I checked it against the published omarchy-4.5.0-3 aarch64 package: all 27 patches apply with fuzz 0, reversing them gives back the pristine files byte for byte, and every patched script and QML file parses. Two details are better than what I had: 0005 summons 4.5's appearance route, and 0012 sends 4.5's absolute-path omarchy-update-restart call to our stand-in. Keeping the 4.0.6 series byte-identical in its own family is exactly right, too.

Some context first: since 6 November our repository carries Omarchy's own 4.0.6 packages (rootfs/packages.pinned, #454), so phones stay on 4.0.6 until the 4.5 series and its companions are ready. That changes what this PR needs to do, and I'd like to split it:

What I'd merge first (no phone test needed while phones are on 4.0.6, since nothing changes for them): the two series families and their selection, plus the migration holds from #469. Before that:

What needs another approach: the update-flow part.

  • The qualification gate requires the exact Omarchy package to be in pacman's cache already, but nothing downloads it (checkupdates runs without -d, and the gate runs before any download). The test puts the file there by hand. On a phone, every update that brings Omarchy would be refused as a whole, even with an approved record. The record also binds the hashes of the updater, the patch tool and both shims, so any later fix would invalidate it. The repository pin now does the holding, so I'd rather not add an approval file. The fuzz-0 replay of the exact package is a good idea; if you keep it, download the package first and hold only Omarchy when it fails.
  • Omarchy 4.0.6 also calls omarchy-update-system-pkgs by name, so the new stand-in would take over every phone's package transaction the day it merges, pin or not. It also drops Omarchy's LC_ALL=C, which the conflict handler needs to read pacman's messages, and when the plan fails it doesn't print pacman's reason. That part needs a run on a Pixel 7 Pro before it can merge, and I can't do one this week.

For the record, here's what still stands between phones and 4.5 outside this PR. I'll track these on #434:

When those are done I'll move the pin to 4.5.0-3 rather than drop it, since the series selection accepts exact versions.

Thanks again; the port itself is most of the work, and it's in good shape.

@marselnenaj
marselnenaj force-pushed the codex/qualified-45-runtime branch from 500585f to 19fb6b1 Compare November 7, 2026 14:20
@marselnenaj marselnenaj changed the title omamobile-omarchy: support the Omarchy 4.5 runtime omamobile-omarchy 0.1.35: select 4.0.6 and 4.5 patch families Nov 7, 2026
@marselnenaj
marselnenaj marked this pull request as ready for review November 7, 2026 14:21
@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj. This is the split I hoped for. The update-flow pieces are all gone and omamobile-update is untouched. The 4.0.6 family is main's series byte for byte (I checked every blob), 0030 stays with jkjoplin, and the Omarchy fixtures are the real v4.5.0 files with the MIT notice shipped. On a 4.0.6 phone the selector picks the same series it runs today, so the hook will just report it as already applied. Thanks also for fixing the sleep-lock extraction: in both 4.0.6 and 4.5 that path is a link to /usr/bin, so the old test was reading the build machine's copy.

Two things before I merge:

  1. The docs that say where the series lives. After this, phones run legacy-4.0.6/, but AGENTS.md (the "Omarchy upgrades overwrite our Omarchy patches" gotcha), README.md (the patches/omarchy/ row) and docs/RELEASES.md §5.x (the "patches survive upgrades" paragraph) still say to edit packages/omamobile-omarchy/patches and make it apply to the Omarchy the phone runs. The next lock or shade fix would then land only in the 4.5 family. A sentence in each would do: two families, picked by the installed Omarchy version, and a fix for today's phones goes in legacy-4.0.6/ (and in patches/ too when it applies to 4.5).
  2. test/lock-glance still has a PRIVATE_LOCK_DIR hook that copies lock files from outside the project, NenajMatrix.qml among them. Please remove it. If there's a lock fix behind it that the series is missing, tell me what it is and we'll bring it in as a patch.

Smaller ones, if you're in there anyway:

  • logic.cjs now names only Omarchy's author, but the reply and forget cases at the end are yours. A line crediting your additions would keep both right.
  • test/sleep-lock.sh applies patches/0021 (now the 4.5 one) to whichever Omarchy package is cached, and on my build machine that's 4.0.6-1. It passes because the two versions differ only by an offset. Choosing the family from the package version, as test/patches.py now does, would make it test what phones run.
  • If the node and QML tests only run when a 4.5 package is present, could nodejs and qt6-declarative stay out of checkdepends? Otherwise a build on a phone pulls in nodejs.
  • The new test files are written as long one-liners. A plainer layout, like the other tests, would help the next person.

Merge order: #469 first, then this rebased onto it, which drops the shared commit. I'll set the versions (0.1.35 and 0.1.36, since #478 took 0.1.34 earlier today) and write the squash message. No phone test is needed while phones stay on 4.0.6. I'll check the patch status on my 7 Pro after installing it, before it's published.

@marselnenaj
marselnenaj force-pushed the codex/qualified-45-runtime branch from 19fb6b1 to f0973e2 Compare November 7, 2026 16:59
@marselnenaj marselnenaj changed the title omamobile-omarchy 0.1.35: select 4.0.6 and 4.5 patch families omamobile-omarchy 0.1.36: select 4.0.6 and 4.5 patch families Nov 7, 2026
angelrenefm added a commit to angelrenefm/omarchy-mobile that referenced this pull request Nov 7, 2026
The case derekross#420's review left open: Hyprland sends a touch to the window
under it whatever the screen's state, so on a dark unlocked phone a tap,
or a double tap, landed in an app no one could see. omamobile-doubletap
now looks at the screen four times a second and, when it goes dark while
unlocked (an idle blank, the panel powered off by hand), runs
omamobile-keys dark-unlocked, which locks behind the dark panel; the
lock's surface then takes the touches. Only where the screen's state can
be told (a DSI connector). The proximity claim's release moves into the
same one-second wait.

README: the dark-unlocked lock, that it works while the phone is awake
(a double tap can't wake it from sleep), and that it times a double tap
from lift to touch while the lock times touch to touch.

omamobile-omarchy 0.1.37, omamobile-shell 0.2.103 (0.1.36 and 0.2.102
are claimed by derekross#471 and derekross#482).

I used AI help (Claude) to work on this.
@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj. Everything from my last two comments is in. The docs in all three places say where a fix goes now, the private lock hook is gone from test/lock-glance, the sleep test picks the family from the package version (and has a negative control), the credit lines are right, and Node and Qt stay out of checkdepends. I checked the legacy family against main again: all 28 files are the same blobs. I also ran the upgrade on a fake root built from the real omarchy-4.0.6-1 aarch64 package: 0.1.35's tool applies main's series, then 0.1.36's tool, with no --series, picks legacy-4.0.6/ and reports it as already applied. A root that claims 4.0.7 is refused, as intended. Making forget return ok for a key that's already gone also matches what the 4.0.6 patch does, which keeps the shell's handling in #468 the same on both.

One thing before I merge, and it's in the 4.5 patch headers. Eight of them (0015, 0016, 0017, 0018, 0019, 0022, 0023 and 0025) have replaced the original header, which says what the patch does and why, with a few lines whose spaces have gone missing, for example in 0023:

# Port mobileglance to4.5auth/wake fields without changing authcapabilities.
# Retain licensed4.0.6 standaloneNotificationCard/NotificationLogic API for
# privatepixel.lock too; parse4.5compactlive/history arrays into legacyreadrows,

The same thing shows in the 4.5 notes of 0017, 0020, 0021, 0024 and 0026 ("Omarchy4.5:"). These headers are the series' documentation, and the text we send upstream, so could you keep each original header and add a short, normally spaced "Omarchy 4.5:" paragraph on what the port changed, as you did well in 0001 and 0012? Please also drop "for private pixel.lock too" from 0023. If the two notification files 0023 brings back (NotificationLogic.js and components/NotificationCard.qml) are only needed by a lock outside this project, please tell me. If the glance needs them itself, a sentence saying so is enough. I should have caught this last time; it was already in the previous push.

Two notes on my side, so you know what happens after the merge. The 4.5 family will start running on phones that already took Omarchy 4.5 before we pinned it, so I'll arrange a first run on one of those before it's published. And #420 adds a lock patch to patches/, which after this PR is the 4.5 family, so whichever of the two is ready first lands first, and the other moves that patch into legacy-4.0.6/. If #420 lands first, please move its 0029 there when you rebase.

I'll write the squash message myself, keeping your Codex co-author line. Thanks again; this is a careful piece of work.

@marselnenaj
marselnenaj force-pushed the codex/qualified-45-runtime branch from f0973e2 to 66b9c17 Compare November 8, 2026 08:30
@marselnenaj

Copy link
Copy Markdown
Contributor Author

Updated in 66b9c17 and rebased onto current main.

Restored the original explanatory headers, preserving their notices, and added short, normally spaced Omarchy 4.5: paragraphs. The remaining 4.5 notes are spaced normally too, and the private-lock wording is removed from 0023.

Both restored notification files are used by the project's own mobile glance: LockGlance.js imports ../notifications/NotificationLogic.js, and LockGlance.qml imports ../notifications/components and instantiates NotificationCard. They are not only dependencies of an external lock.

All functional patch payloads and the entire legacy family remain byte-identical to the previous PR head. The rebase preserves main's new remote-testing documentation alongside the patch-family guidance. Header invariance checks, tools/pr-check and tests/packages omamobile-omarchy against the actual Omarchy 4.5.0-3 aarch64 package pass.

#420 is still open, so no 0029 move was needed. If it lands first, its family placement will need to be handled when rebasing, as you described. Human authorship and the Codex co-author trailers are retained.

derekross added a commit that referenced this pull request Nov 8, 2026
Thirty-one lessons from round 44's reviews, confirmed by the maintainer, plus a second instance of L-454 from #493. #420's and #471's reviews found the same patch-series lesson from either side; it is one entry (L-550).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj. The headers read well now. Each of the eight has its original explanation back, word for word, with a short "Omarchy 4.5:" paragraph under it, and the private-lock wording is gone. Thanks also for the answer about NotificationLogic.js and NotificationCard; 0023's note now says the glance uses them. I checked that the patch bodies didn't change: from the first diff line on, all 26 files are byte for byte what I reviewed last time, and the 28 files in legacy-4.0.6/ are still main's series, blob for blob.

This time I also ran the package's test/all against the signed omarchy-4.5.0-3 aarch64 package from Omarchy's stable repository, and against 4.0.6-1. Both pass with nothing skipped. On 4.5 all 26 patches apply at fuzz 0, the patched QML parses, the re-apply is idempotent, and 4.6.0 is refused. On 4.0.6, 28 patches apply.

A few small things. I can fix them at merge if you'd rather not push again:

  • 0027 and 0028 still have the squashed spacing in their last line: # Omarchy4.5: local transport-only media controls. and # Omarchy4.5: mobile narrowterminal logo sizing. My list last time missed those two, sorry.
  • test/mobile-pam-preserve.py isn't called by anything, so it never runs. I ran it by hand against 4.5.0-3's omarchy-apply-lock with 0022 applied, and it passes. Calling it from port45.actual() after the apply would make it count.
  • The default_series docstring still says "never apply4.5 contexts", and the README's patches/omarchy/ row still mentions the bridge-* stopgaps, which now live only in legacy-4.0.6/.

Before I can merge, one run on a phone. The development channel is published from main every morning, so once this merges, a phone that is already on Omarchy 4.5.0-3 gets the 4.5 family at its next update. That has never run on a phone. Could you do that run on your phone, if it's on 4.5.0-3, with this head's package installed and no private changes to Omarchy's files? If there are some, list them. The tool refuses to patch over a changed file, and that's worth knowing too. Then please post:

  1. omamobile-omarchy-patches status: I'd expect 26 applied and the series /usr/share/omamobile/omarchy/patches.
  2. Lock the phone, then unlock it with the PIN pad. Open the notification shade.
  3. omamobile-update check --json (read-only, no update).
  4. The tools/phone-smoke report.

If you can't, I'll ask someone with a 4.5 phone, or do it myself when I'm back. On my 7 Pro, which stays on 4.0.6, I'll check the patch status after installing it.

On #420: I tried its 0029 inside your 4.5 family on the real package, and it applies cleanly there too. So the order doesn't break anything at apply time. Whichever lands second still moves 0029 into legacy-4.0.6/, so that 4.0.6 phones keep it.

I'll write the squash message myself and keep your Codex co-author line. Thanks again for the patience on this one.

@marselnenaj

marselnenaj commented Nov 8, 2026 •

Copy link
Copy Markdown
Contributor Author

Thanks for checking both signed package versions. I have corrected the remaining header/doc spacing locally and wired mobile-pam-preserve.py into port45.actual() after apply. The real 4.5 package source suite passes, including that behavioral check, with no skips. Pushed as ee8b926. The complete package suite also passes against both real 4.5.0-3 and 4.0.6-1 packages without skips; all 26 runtime patch bodies and the legacy family remain unchanged.

I cannot supply the requested clean-phone qualification from my current development phone: it runs 4.5.0-3 with additional shell, notification, lock and input adaptations. I will not present that as an unmodified installation or remove them just for this test. A clean 4.5 device run is still needed before merge; the source tests do not replace your PIN/unlock and phone-smoke checklist. I can prepare a sanitized changed-file inventory separately.

Understood on #420: whichever lands second must retain 0029 in legacy-4.0.6. Thank you for preserving the attribution.

@jkjoplin

jkjoplin commented Nov 8, 2026

Copy link
Copy Markdown
Contributor

I ran this on my Pixel 7 Pro (cheetah). It has been on Omarchy 4.5.0-3 since 2026-11-05. I built this head's package (ee8b926) on the phone with tools/phone-install-package packages/omamobile-omarchy and installed it over a local 0.1.33-0 build (#431's content).

The phone before the install

  • pacman -Qkk omarchy omarchy-settings: 0 altered files (of 2329 and 1093). Omarchy's files have no private changes.
  • It took 4.5.0-3 through omarchy update on 2026-11-05, before the pin. The patch hook failed at 0001 then (applied: []), and the update stopped at migration 1788163635 (the sudo grants). So 73 of 4.5.0-3's 218 migrations haven't run, 1791403310 (titlebars) among them. I didn't mark any migration done by hand.
  • After the battery went flat on 11-07, the phone booted into unpatched 4.5.0-3, and it ran that way until this install.

1. omamobile-omarchy-patches status: result: ok, 26 applied, 0 already in Omarchy, 24 files changed, series /usr/share/omamobile/omarchy/patches. The hook applied them during the install; then omarchy-restart-shell. I haven't rebooted.

2. Lock, PIN, notifications: I locked the phone with the power key and unlocked it with the PIN pad, with no problem. The journal shows omarchy-lock-password authenticating and the lock logging unlocked. A critical toast swipes away (closed_reason: dismissed), and a finger tap on the close glyph closes one too (dismissed, no action run). Normal toasts expire after 8 s, as 4.5 intends.

0017's 4.5 port works, but its swipe animation is rough. Past the threshold, the card follows the finger sideways, then jumps back to its place and leaves upward with the scene's fade. I saw it twice. The cause is onActiveChanged: it sets card.swipeOffset = 0, with no animation, before it calls card.dismissed(), so the leave starts from the card's resting place. A swipe short of the threshold snaps back the same way. The legacy 0017 animates the offset off the side (to ±(width + Style.space(24)) in 140 ms, InQuad), fades the card with distance, takes a flick past 10%, and springs a short swipe back (200 ms, OutBack). The 4.5 header still describes that behavior (35% or a flick, then it flies off), but the port uses min(Style.space(90), 25%) and no flick. Carrying the legacy animation over, and calling dismissed() when the fly-off ends, would fix it. That could just as well be a follow-up.

The shade itself isn't right yet, but that's #482's part, not this PR's. omamobile-shell 0.2.100 still reads 4.0.6's per-file store (notifications/history/*.json), so it showed 7 notifications from before the update and none of the new ones. Clear all didn't remove them either: they came back, because the 4.5 service's clear only empties history.json. I moved the 7 old files aside.

3. omamobile-update check --json: ok: true, no blockers, omarchy_update empty, patches ok with all 26 listed, and 12 ordinary updates. I didn't update.

omamobile-update check --json
{
    "battery": {
        "charging": true,
        "percent": 43,
        "present": true
    },
    "battery_ok": true,
    "blockers": [],
    "boot_packages": [],
    "channel": {
        "error": "",
        "name": "development",
        "server": "https://pkgs.omamobile.org/development/$arch"
    },
    "checked_offline": false,
    "credential": "pin",
    "free_gib": 93.9,
    "kernel": null,
    "min_battery": 30,
    "ok": true,
    "omarchy": "4.5.0-3",
    "omarchy_update": "",
    "patches": {
        "applied": [
            "0001-first-run-let-a-platform-skip-the-keybindings-welcome.patch",
            "0003-lock-number-pad-for-touch-only-devices.patch",
            "0005-touch-swipe-the-image-picker-and-hold-the-desktop-for-style.patch",
            "0006-screensaver-restore-the-cursor-and-hide-the-on-screen-keyboard.patch",
            "0007-idle-lock-when-the-user-returns-to-the-screensaver.patch",
            "0008-brightness-panels-with-bl-power-follow-dpms.patch",
            "0009-lock-ipc-wake-for-hardware-keys.patch",
            "0010-polkit-leave-the-on-screen-keyboard-usable.patch",
            "0011-lock-and-polkit-pin-or-password-input.patch",
            "0012-update-hand-omarchy-update-to-the-phone-wrapper.patch",
            "0013-idle-wake-a-locked-display-through-the-lock.patch",
            "0014-lock-emergency-call-button-for-touch-devices.patch",
            "0015-notifications-inline-reply.patch",
            "0016-pacman-a-phone-refuses-the-channel-switch-and-pacman-refresh.patch",
            "0017-notifications-swipe-to-dismiss-and-a-touch-close-button.patch",
            "0018-menu-a-back-button-and-swipe-right-to-go-back.patch",
            "0019-notifications-forget-one-notification-by-ipc.patch",
            "0020-notifications-touch-phrasing-for-keyboard-and-click-hints.patch",
            "0021-sleep-turn-the-display-off-before-suspend.patch",
            "0022-lock-fingerprint-only-when-a-finger-is-enrolled.patch",
            "0023-lock-a-glance-view-for-touch-only-devices.patch",
            "0024-app-icons-keep-the-largest-size.patch",
            "0025-launch-no-toast-for-shell-surfaces.patch",
            "0026-lock-battery-status.patch",
            "0027-lock-media-controls.patch",
            "0028-show-logo-a-stacked-logo-for-a-narrow-terminal.patch"
        ],
        "failed": null,
        "log": "/var/log/omamobile/omarchy-patches.log",
        "message": "26 applied, 0 already in Omarchy; 24 files changed",
        "omarchy": "omarchy 4.5.0-3",
        "result": "ok",
        "time": "2026-11-08T14:00:53-07:00",
        "upstream": []
    },
    "release_notes": "https://github.com/basecamp/omarchy/releases/latest",
    "repositories": "",
    "repositories_ok": true,
    "running": false,
    "update_count": 12,
    "updates": [
        "fastfetch 2.70.0-1 -> 2.71.0-1",
        "frei0r-plugins 3.6.0-1 -> 3.7.0-1",
        "grok-bot 0.79.2-1 -> 0.79.3-1",
        "inotify-tools 4.26.270-1 -> 4.26.272-1",
        "inxi 3.3.41.1-1 -> 3.3.41.1-2",
        "kaccounts-integration 26.08.2-1 -> 26.08.3-1",
        "kdenlive 26.08.2-3 -> 26.08.3-1",
        "libspelling 0.4.10-2 -> 0.4.11-1",
        "mise-bin 2026.11.0-1 -> 2026.11.4-1",
        "omamobile-audio 0.5.13-1 -> 0.5.15-1",
        "signon-kwallet-extension 26.08.2-1 -> 26.08.3-1",
        "usage 6.12.1-1 -> 7.3.0-1"
    ]
}

4. tools/phone-smoke: identical to a run just before the install, apart from the omamobile-omarchy version. All required checks are ok. The "new dmesg errors" row is the same wc-mbox and cpif boot-time noise as before the install.

phone-smoke report

phone-smoke: cheetah

Kernel 7.3.0-rc6-omamobile-push1-g1cb4bd36f3be, tools at ee8b926, 2026-11-08T21:02Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-omamobile-push1-g1cb4bd36f3be, slot b, up 674s
storage guard (required) ok 55 partitions; only userdata and slot B's boot images writable
root (required) ok ext4 on userdata, 94G free
kernel health (required) ok no oops or BUG; 0 WARN, 399 error lines in dmesg (57 known, of them ASoC no backend DAIs (PipeWire probe) x1, 342 new)
new dmesg errors not ok wc-mbox N: Try increasing MBOX_TX_QUEUE_LEN x282; cpif: rx_demux: PCIE: ERR! N is not opened x15; cpif: pass_skb_to_demux: PCIE: N<N: ERR! N>recv_skb fail (N) x15
systemd units ok none failed
USB gadget (required) ok bound to 11210000.usb
display ok card2-DSI-1 connected, 1440x3120
GPU ok render node on panthor
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present, the shell can lock (passwordPam true, locked false)
touch ok Synaptics TouchComm Touchscreen
buttons ok gpio-keys
Wi-Fi ok wlp1s0 on brcmfmac, link up
Bluetooth ok hci0, powered false
battery ok 44% Charging, 39 °C
sound cards ok googleaocsndcar
sensors ok Accelerometer AmbientLight Proximity
cameras ok 4 listed by libcamera
thermal ok hottest zone 39 °C

Packages: hyprland 0.57.0-3.1, libcamera 0.7.2-4.22, mesa 1:26.2.4-1.4, omacam 0.3.3-1, omamessages 0.2.2-1, omamobile-apps 0.2.2-1, omamobile-audio 0.5.13-1, omamobile-device-cheetah 0.1.44-1, omamobile-fingerprint 0.1.15-1, omamobile-firstboot 0.1.3-1, omamobile-gnss 0.2.7-1, omamobile-keyring 20261026-1, omamobile-mirror 0.1.1-1, omamobile-modem 0.1.16-1, omamobile-omarchy 0.1.36-1, omamobile-owl-widgets 0.1.2-1, omamobile-sensors 0.2.1-1, omamobile-shell 0.2.100-1, omaphone 0.2.3-1, omarchy 4.5.0-3, omatel-agent 0.2.4-2, quickshell 0.3.2.r7.g2b4501a-1

If you'd like anything else from this phone (a reboot, or a particular patch exercised), let me know. It stays on 4.5.0-3 with this build until the series ships.

@marselnenaj

Copy link
Copy Markdown
Contributor Author

Thank you for testing the exact ee8b926 package on cheetah and documenting the before/after state. The successful PIN unlock, 26-patch application, update check and unchanged required smoke results are valuable device evidence. I will retain your qualification that 73 migrations have not run; this is not evidence that the complete 4.5 migration sequence has finished.

The 0017 offset reset and missing fly-off/snap-back behavior are concrete regressions. I am preparing a small follow-up that restores the gesture animation and defers dismissal until the fly-off completes, with Qt gesture regression tests. I will credit your report and keep it separate from this already-tested compatibility head.

The shade storage mismatch is covered separately by #482. The snapshot-delete provider follow-up is linked above, and #533 waits for daemon deletion acknowledgment so failed deletion does not silently discard rows or hide new arrivals. Both remain separate from this compatibility PR.

I do not need you to reboot or run further hardware probes for the animation work. I will link the reviewed follow-up once it is ready.

@marselnenaj

Copy link
Copy Markdown
Contributor Author

The separate animation follow-up is now published:

Thanks to @jkjoplin for the precise device report. The follow-up restores the fly-off before dismissal, short-swipe spring-back, directional flick and scene-composed fading, and cancels interrupted drags safely. The genuine 4.5 package gate passes; 27 Qt entries have no failures/skips, the parent baseline and four inverse controls detect the regressions, and all seven policy gates pass. Independent review approved the exact final head. The 28 legacy-family files are unchanged.

I have not installed this animation change on a phone. Your earlier device test qualifies the preceding ee8 compatibility build, not this follow-up. Version 0.1.37 is provisional for allocation at merge.

I am linking the small follow-up here while #471 is pending rather than opening a PR that repeats its parent. After #471 merges, I will create a new branch from current upstream main and cherry-pick only these follow-ups, leaving this published branch unchanged.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj, for the quick fixes: 0027 and 0028 read normally, the docstring and README are right, and mobile-pam-preserve.py now runs from port45.actual() (I saw it pass against the real 4.5.0-3 package). And thank you, @jkjoplin, for the 4.5.0-3 run on your 7 Pro. That was the run this PR was waiting for, and the details (no altered Omarchy files, 26 applied, the PIN unlock, the update check, and the 73 migrations that haven't run) are exactly what I needed. The 0017 swipe animation you found is a real regression; since dismissal itself works, I'm happy for Marsel's fix to come as its own PR after this one, with a run on a 4.5 phone.

One thing has changed under the PR: #420 landed overnight as omamobile-omarchy 0.1.37, with its double-tap lock patch as patches/0029-lock-double-tap-to-wake-for-touch-only-devices.patch. In a plain merge that file lands in the 4.5 family, so legacy-4.0.6/ stays at 28 and every phone on 4.0.6, the release phones included, would lose double tap to wake at their next update. I built the merge three ways and ran test/all against the signed 4.0.6-1 and 4.5.0-3 packages:

  • plain merge: on 4.0.6 only 28 apply, and the 0029 check and four of the glance's QML tests fail. CI wouldn't notice, because it has no Omarchy package;
  • 0029 moved into legacy-4.0.6/: 29 apply on 4.0.6 and 26 on 4.5, and everything passes except test/source-family.py, which expects 28 legacy patches.

So could you rebase onto main with:

  1. git mv packages/omamobile-omarchy/patches/0029-lock-double-tap-to-wake-for-touch-only-devices.patch packages/omamobile-omarchy/legacy-4.0.6/ (a move, not a copy, so the 4.5 family stays exactly what jkjoplin tested; 0029 on 4.5 can come later with its own run);
  2. test/source-family.py: == 28 to == 29;
  3. the version at 0.1.39, since 0.1.37 is on main and my omamobile-omarchy 0.1.38: omamobile-update refuses to run as root, and passes OMARCHY_PATH on #548 takes 0.1.38, and the title to match;
  4. if you like, the module docstring at the top of omamobile-omarchy-patches, which still names only patches/.

If you'd rather not push again, tell me and I'll make exactly those changes at merge. On a fake root from the real 4.0.6-1 package, upgrading from main's 0.1.37 to that result reports "series already applied" with 29 patches from legacy-4.0.6/, so phones on 4.0.6 shouldn't see a change. I'll still check that on my 7 Pro before it's published.

On timing: I'll merge it right after the 2026.11.08 release is out and #548 is in. I'll write the squash message myself and keep your Codex co-author line. Thanks again, both of you.

derekross added a commit that referenced this pull request Nov 9, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (#547, #548, #549,
#550, #560). Repeats of existing lessons extend them instead: L-45
(kernel #222), L-183 (#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (#471 with #420). C-34 and C-35
correct our #290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on #551 (round 45's lessons).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross added a commit that referenced this pull request Nov 9, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (#547, #548, #549,
#550, #560). Repeats of existing lessons extend them instead: L-45
(kernel #222), L-183 (#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (#471 with #420). C-34 and C-35
correct our #290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on #551 (round 45's lessons).

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
marselnenaj and others added 5 commits November 9, 2026 20:49
Keep the 4.0.6 series byte-identical while porting runtime contexts to
4.5.0. Select only supported source families and include mobile migration
holds. Preserve the repository pin and production updater behavior.

Do not ship the new updater gate, transaction shim or approval records.
Keep the merged PR431 in the legacy archive byte for byte and leave its
4.5 port to jkjoplin. Preserve upstream
copyright and MIT notices; remove false fixture-extension attribution.

Run the full adapter suite against the signed 4.5.0-3 package, including
fuzz-zero apply/reverse, idempotence, source selection, notification
functions, 16 QML parses and native lock fixtures. No device update or
physical smoke result is claimed.

Co-Authored-By: Codex <codex@openai.com>
Document the version-selected patch families so fixes for current 4.0.6
phones reach the legacy family. Remove the external private-lock fixture
hook; the suite uses only packaged lock sources and repository fixtures.
Credit the mobile fixture additions to their confirmed human author while
retaining Omarchy's original notice.

Choose the sleep patch from the actual package version, keep Node and QML
test tools optional, and format the new fixtures for maintenance. Make
forget-by-key idempotent for valid missing keys and assert a repeated
delete leaves state unchanged; malformed keys still fail.

Run full adapter suites against the hash-pinned 4.0.6-1 and signed 4.5.0-3
packages, plus four inert family-selection and negative-control cases.
Both suites pass. No external lock assets, phone actions or authentication
policy changes are involved.

Co-Authored-By: Codex <codex@openai.com>
Restore the original explanatory headers and retain their notices. Describe
4.5-specific changes in normally spaced paragraphs so the series remains
useful documentation when fixes are maintained or sent upstream.

Remove the private-lock wording from 0023. The restored NotificationLogic
and NotificationCard APIs are used by the project mobile glance itself.
Every functional patch payload and the entire legacy family remain identical.
Header checks and the full actual-package 4.5 adapter suite pass.

Co-Authored-By: Codex <codex@openai.com>
Exercise the actual patched PAM guard in the real-package rehearsal instead of leaving its behavioral fixture uncalled. Keep the retained legacy stopgaps explicit in documentation and fix the remaining patch-header spacing.

Co-Authored-By: Codex <codex@openai.com>
Rebase the version-selected series on current main and retain its new double-tap-to-wake patch in the 4.0.6 family. Move the patch without changing its contents, expect all 29 legacy patches, and describe both installed source families in the tool docstring.

Co-Authored-By: Codex <codex@openai.com>
@marselnenaj
marselnenaj force-pushed the codex/qualified-45-runtime branch from ee8b926 to 95a536a Compare November 9, 2026 20:03
@marselnenaj marselnenaj changed the title omamobile-omarchy 0.1.36: select 4.0.6 and 4.5 patch families omamobile-omarchy 0.1.39: select 4.0.6 and 4.5 patch families Nov 9, 2026
@marselnenaj

Copy link
Copy Markdown
Contributor Author

Rebased onto current main (b5207b86) and bumped the package to 0.1.39. I moved 0029 into legacy-4.0.6/, updated the assertion to 29, and corrected the tool's module docstring to describe both families.

All 29 legacy patches are byte-identical to main's original flat series, including the double-tap patch; the 26-patch 4.5 family is byte-identical to the previous PR head. Both complete package suites passed against the cached real 4.0.6-1 and 4.5.0-3 archives. The legacy Qt suite passed 56 tests without skips, the touched QML parsed for both families, and the commit-range policy checks passed.

The updater, #486 migration holds, package pins and credits are retained unchanged. I kept the 0017 animation work separate. This revision has not been installed on a phone; your planned 4.0.6 phone check is still needed, and I am not treating the archive rehearsals as execution of all pending migrations.

@derekross

Copy link
Copy Markdown
Owner

Thank you, @marselnenaj. All four points are in, and I checked each one against the new head:

  • 0029 is a pure rename into legacy-4.0.6/, and the 29 files there are byte for byte main's series. The 4.5 family is the same tree as the head @jkjoplin ran on his phone, so his run still stands.
  • test/all against the signed packages: on 4.0.6-1, 29 applied, the double-tap check passes and the glance's Qt suite is 56 of 56; on 4.5.0-3, 26 applied and the family, PAM and unknown-version checks pass. Nothing skipped in either run.
  • I also rehearsed the upgrade a release phone will see: a root built from the real 4.0.6-1 package with the published 0.1.37 installed and applied, then this head on top. The hook reports "series already applied" from legacy-4.0.6/ and not one of Omarchy's files changes.

What's left is mine: I'll install it on my Pixel 7 Pro, read omamobile-omarchy-patches status and verify (29 applied, series legacy-4.0.6), and merge. I'll write the squash message myself and keep your Codex co-author line.

For #585 afterwards: this lands as a squash, so git rebase --onto origin/main 95a536a3 <your branch> will carry only the provider commit across.

One small thing for a later PR, not this one: the 4.0.6 real-package case in test/patches.py passes --series itself, so the selector never runs against a real 4.0.6 root in the tests. One apply without --series there would cover it.

@derekross
derekross merged commit 58863fa into derekross:main Nov 10, 2026
2 checks passed
@derekross

Copy link
Copy Markdown
Owner

Merged as omamobile-omarchy 0.1.39. On my Pixel 7 Pro (Omarchy 4.0.6, 0.1.37 installed) the upgrade did what the rehearsal said: status and verify report 29 patches from legacy-4.0.6, 0029 among them, every file under /usr/share/omarchy kept its checksum, the log says "series already applied", and the shell didn't restart. Thank you, @marselnenaj, and thank you @jkjoplin for the 4.5 run.

derekross added a commit that referenced this pull request Nov 11, 2026
…ification store

The review of #471 found that the bar's unread count and the quick-settings
shade read Omarchy 4.0's notification files (one <timestamp>-<id>.json per
notification under ~/.local/state/omarchy/notifications/ and history/),
which Omarchy 4.5 no longer writes: its Store.js keeps live.json (toasts on
screen) and history.json (closed rows, newest first, 100 over 24 hours), one
JSON array each, rows named by a key with `ts` in seconds, and its migration
1792648436 deletes the 4.0 files. On 4.5 the shade would show nothing, the
badge would never count, and Clear all would call a `clear` that 4.5 lacks.

Both layouts are now read, chosen by what exists, not by version: where
history.json exists the shade reads live.json and history.json, and Clear
all is dismissAll followed by `forget <key>` for every notification read
(the 4.5 version of patch 0019 in #471 takes the key); the bar counts
history.json rows newer than the shade's mark, watching the file and
re-reading it every 30 s, since the service replaces it with rename(2).
Without history.json, every command and count is the one 0.2.101 runs.

Tested: test/all (the offscreen touch tests included); new rows in
test/notifications-model.js and test/status-model.js in 4.5.0-3's format,
made by its own Store.js; test/notifications-store.sh runs the shade's read
and Clear all commands from QuickSettings.qml on both layouts with a fake qs;
test/quicksettings-render now draws the light theme over 4.5 files and
checks that all three rows show (run in a sandbox like the OWL container,
with Omarchy 4.0.6's shell: 3 rows on both layouts; main's plugin shows 0 on
4.5). Mutations of each new branch fail a test. Not run on a phone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross added a commit that referenced this pull request Nov 11, 2026
…ification store

The review of #471 found that the bar's unread count and the quick-settings
shade read Omarchy 4.0's notification files (one <timestamp>-<id>.json per
notification under ~/.local/state/omarchy/notifications/ and history/),
which Omarchy 4.5 no longer writes: its Store.js keeps live.json (toasts on
screen) and history.json (closed rows, newest first, 100 over 24 hours), one
JSON array each, rows named by a key with `ts` in seconds, and its migration
1792648436 deletes the 4.0 files. On 4.5 the shade would show nothing, the
badge would never count, and Clear all would call a `clear` that 4.5 lacks.

Both layouts are now read, chosen by what exists, not by version: where
history.json exists the shade reads live.json and history.json, and Clear
all is dismissAll followed by `forget <key>` for every notification read
(the 4.5 version of patch 0019 in #471 takes the key); the bar counts
history.json rows newer than the shade's mark, watching the file and
re-reading it every 30 s, since the service replaces it with rename(2).
Without history.json, every command and count is the one 0.2.101 runs.

Tested: test/all (the offscreen touch tests included); new rows in
test/notifications-model.js and test/status-model.js in 4.5.0-3's format,
made by its own Store.js; test/notifications-store.sh runs the shade's read
and Clear all commands from QuickSettings.qml on both layouts with a fake qs;
test/quicksettings-render now draws the light theme over 4.5 files and
checks that all three rows show (run in a sandbox like the OWL container,
with Omarchy 4.0.6's shell: 3 rows on both layouts; main's plugin shows 0 on
4.5). Mutations of each new branch fail a test. Not run on a phone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross added a commit that referenced this pull request Nov 11, 2026
…ification store

The review of #471 found that the bar's unread count and the quick-settings
shade read Omarchy 4.0's notification files (one <timestamp>-<id>.json per
notification under ~/.local/state/omarchy/notifications/ and history/),
which Omarchy 4.5 no longer writes: its Store.js keeps live.json (toasts on
screen) and history.json (closed rows, newest first, 100 over 24 hours), one
JSON array each, rows named by a key with `ts` in seconds, and its migration
1792648436 deletes the 4.0 files. On 4.5 the shade would show nothing, the
badge would never count, and Clear all would call a `clear` that 4.5 lacks.

Both layouts are now read, chosen by what exists, not by version: where
history.json exists the shade reads live.json and history.json, and Clear
all is dismissAll followed by `forget <key>` for every notification read
(the 4.5 version of patch 0019 in #471 takes the key); the bar counts
history.json rows newer than the shade's mark, watching the file and
re-reading it every 30 s, since the service replaces it with rename(2).
Without history.json, every command and count is the one 0.2.101 runs.

Tested: test/all (the offscreen touch tests included); new rows in
test/notifications-model.js and test/status-model.js in 4.5.0-3's format,
made by its own Store.js; test/notifications-store.sh runs the shade's read
and Clear all commands from QuickSettings.qml on both layouts with a fake qs;
test/quicksettings-render now draws the light theme over 4.5 files and
checks that all three rows show (run in a sandbox like the OWL container,
with Omarchy 4.0.6's shell: 3 rows on both layouts; main's plugin shows 0 on
4.5). Mutations of each new branch fail a test. Not run on a phone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross added a commit that referenced this pull request Nov 11, 2026
…ification store (#482)

The review of #471 found that the bar's unread count and the quick-settings
shade read Omarchy 4.0's notification files (one <timestamp>-<id>.json per
notification under ~/.local/state/omarchy/notifications/ and history/),
which Omarchy 4.5 no longer writes: its Store.js keeps live.json (toasts on
screen) and history.json (closed rows, newest first, 100 over 24 hours), one
JSON array each, rows named by a key with `ts` in seconds, and its migration
1792648436 deletes the 4.0 files. On 4.5 the shade would show nothing, the
badge would never count, and Clear all would call a `clear` that 4.5 lacks.

Both layouts are now read, chosen by what exists, not by version: where
history.json exists the shade reads live.json and history.json, and Clear
all is dismissAll followed by `forget <key>` for every notification read
(the 4.5 version of patch 0019 in #471 takes the key); the bar counts
history.json rows newer than the shade's mark, watching the file and
re-reading it every 30 s, since the service replaces it with rename(2).
Without history.json, every command and count is the one 0.2.101 runs.

Tested: test/all (the offscreen touch tests included); new rows in
test/notifications-model.js and test/status-model.js in 4.5.0-3's format,
made by its own Store.js; test/notifications-store.sh runs the shade's read
and Clear all commands from QuickSettings.qml on both layouts with a fake qs;
test/quicksettings-render now draws the light theme over 4.5 files and
checks that all three rows show (run in a sandbox like the OWL container,
with Omarchy 4.0.6's shell: 3 rows on both layouts; main's plugin shows 0 on
4.5). Mutations of each new branch fail a test. Not run on a phone.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
Thirty-one lessons from round 44's reviews, confirmed by the maintainer, plus a second instance of L-454 from derekross#493. derekross#420's and derekross#471's reviews found the same patch-series lesson from either side; it is one entry (L-550).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
Round 46's confirmed candidates from the reviews of the open
omarchy-mobile and linux-omarchy-mobile pull requests, the round's own
process slips (shared refs and scratch files deleted by glob, a
maintainer fixup that skipped the package's test/all, an amend on a
contributor branch) and the 2026.11.08 release PRs (derekross#547, derekross#548, derekross#549,
derekross#550, derekross#560). Repeats of existing lessons extend them instead: L-45
(kernel derekross#222), L-183 (derekross#409), L-426 (the round's version collisions,
now set by hand at merge) and L-550 (derekross#471 with derekross#420). C-34 and C-35
correct our derekross#290 shell suggestion and round 45's claim that
tests/dipper-probe guards anything.

Stacked on derekross#551 (round 45's lessons).

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
…oss#471)

Omarchy 4.5 changed the files most of our patches touch, so one series
no longer fits both. omamobile-omarchy now carries two families and picks
by the installed Omarchy version: patches/ is the 26-patch port to
Omarchy 4.5.0, and legacy-4.0.6/ is the 4.0.6 series as it was, 29
patches with derekross#420's double tap to wake (0029). Any other Omarchy version
is refused, so a new Omarchy never gets a half-applied series.

Checked against the signed aarch64 packages: 29 apply on omarchy 4.0.6-1
and 26 on 4.5.0-3, each with --fuzz=0, and the package's tests pass on
both. The 4.5 family ran on a contributor's Pixel 7 Pro on Omarchy
4.5.0-3. On the maintainer's Pixel 7 Pro (Omarchy 4.0.6, 0.1.37
installed), the upgrade to 0.1.39 on 2026-11-09 rewrote no Omarchy file:
status and verify report 29 patches from legacy-4.0.6, the log says
"series already applied", and the shell didn't restart.

Co-Authored-By: Codex <codex@openai.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
…ification store (derekross#482)

The review of derekross#471 found that the bar's unread count and the quick-settings
shade read Omarchy 4.0's notification files (one <timestamp>-<id>.json per
notification under ~/.local/state/omarchy/notifications/ and history/),
which Omarchy 4.5 no longer writes: its Store.js keeps live.json (toasts on
screen) and history.json (closed rows, newest first, 100 over 24 hours), one
JSON array each, rows named by a key with `ts` in seconds, and its migration
1792648436 deletes the 4.0 files. On 4.5 the shade would show nothing, the
badge would never count, and Clear all would call a `clear` that 4.5 lacks.

Both layouts are now read, chosen by what exists, not by version: where
history.json exists the shade reads live.json and history.json, and Clear
all is dismissAll followed by `forget <key>` for every notification read
(the 4.5 version of patch 0019 in derekross#471 takes the key); the bar counts
history.json rows newer than the shade's mark, watching the file and
re-reading it every 30 s, since the service replaces it with rename(2).
Without history.json, every command and count is the one 0.2.101 runs.

Tested: test/all (the offscreen touch tests included); new rows in
test/notifications-model.js and test/status-model.js in 4.5.0-3's format,
made by its own Store.js; test/notifications-store.sh runs the shade's read
and Clear all commands from QuickSettings.qml on both layouts with a fake qs;
test/quicksettings-render now draws the light theme over 4.5 files and
checks that all three rows show (run in a sandbox like the OWL container,
with Omarchy 4.0.6's shell: 3 rows on both layouts; main's plugin shows 0 on
4.5). Mutations of each new branch fail a test. Not run on a phone.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants